RPV vs AOV
RPV measures revenue per visitor; AOV measures revenue per order. They move independently — and reading them together is how you tell a winning test from a vanity one.
RPV vs AOV
RPV (revenue ÷ visitors) measures how much each session is worth; AOV (revenue ÷ orders) measures how much each buyer spends.
Revenue per visitor and average order value sit one denominator apart, but they answer completely different questions. RPV asks: of everyone who landed on the site, how much money did we extract on average? AOV asks: of the people who actually bought, how big was their basket?
Because RPV bakes conversion rate into its numerator and AOV ignores it entirely, the two metrics can — and routinely do — move in opposite directions during the same test. A bundle offer can lift AOV by 18% while RPV stays flat. A checkout simplification can lift RPV by 9% while AOV drops. Reading them together is how you tell which kind of win you actually got.
The math makes the relationship explicit: RPV = conversion rate × AOV. That single identity is why the two metrics drift apart. Anything that changes who buys (conversion rate) shows up in RPV but not AOV. Anything that changes what buyers spend (basket size) shows up in both — usually.
The "usually" is the trap. A free-shipping threshold at €75 can raise AOV by pulling small baskets up, while simultaneously scaring off price-sensitive visitors who would have converted at €40. AOV goes up. Conversion rate goes down. RPV — the only number tied to actual revenue — barely moves, or moves negative.
How RPV and AOV decouple across common test types
| Test scenario | Conversion rate Δ | AOV Δ | RPV Δ | Real win? |
|---|---|---|---|---|
| Simplified one-page checkout (apparel) | +12% | −3% | +8.6% | Yes — RPV up |
| Free-shipping threshold raised €50 → €75 | −6% | +15% | +8.1% | Yes, but check margin |
| Cross-sell modal on cart page | −2% | +9% | +6.8% | Yes — RPV up |
| Hero discount banner (10% off) | +18% | −9% | +7.4% | Yes — but discount eats margin |
| Premium bundle as default selection | −11% | +22% | +8.6% | Yes — but watch repeat rate |
| Trust badges added to PDP | +5% | +0% | +5.0% | Yes — clean RPV win |
Notice the pattern: every row that looks like a winner on AOV alone has a conversion-rate cost, and every row that looks like a winner on conversion rate alone has an AOV cost. RPV is the only column where the trade-off has already been resolved into a single number that maps directly to revenue.
When each metric is the right scoreboard
Use RPV as the primary metric for almost every site-wide CRO test: landing-page redesigns, checkout flows, PDP layouts, navigation changes, trust signals. These touch the whole funnel, so the only fair scoreboard is per-visitor revenue. AOV alone will mislead you whenever the test affects who buys, not just what they buy.
Use AOV as the primary metric only for tests that are structurally downstream of the purchase decision — post-add-to-cart upsells, order-bump checkboxes, bundle defaults on the cart page, shipping-threshold tweaks. These tests assume the visitor already decided to buy; the question is just basket size. Even then, watch conversion rate as a guardrail.
The AOV trap
If a test lifts AOV but drops conversion rate, AOV is doing the wrong job: it's filtering for richer baskets by losing the rest. The shipping-threshold example above is the canonical case — AOV up 15%, conversion rate down 6%, RPV barely positive. Teams that report on AOV alone ship this test and quietly lose revenue for a quarter before anyone notices.
Reading them together: the 2×2
The cleanest diagnostic is to plot the directional change in conversion rate against the directional change in AOV after a test. Four quadrants, four very different stories. Same-direction wins (both up) are unambiguous. Opposite-direction results need RPV — and a margin check — to call.
On a checkout test for a Shopify apparel store with €60 baseline AOV and 2.5% baseline conversion, a +10% RPV lift is roughly €1.50 more per visitor. At 80,000 monthly sessions that's €120,000 in annualised revenue from one ship — which is why getting the scoreboard right matters more than the test idea itself. The RPV drivers framework breaks this down further.
RPV impact across four test outcomes (apparel store, €60 AOV, 2.5% CVR baseline)
RPV vs AOV: frequently asked questions
RPV = conversion rate × AOV. If 100 visitors land on a page, 3 buy, and average order is €80, then AOV is €80, conversion rate is 3%, and RPV is €2.40. Change any one of those and the relationship updates.
RPV for almost everything. It's the only metric that captures conversion rate and basket size in one number, which means it can't be gamed by a test that lifts one at the expense of the other. Use AOV only for tests that are strictly post-purchase-decision (upsells, bundles, order bumps).
Yes, and it's common. A discount banner that lifts conversion rate by 18% while dropping AOV by 9% still produces a +7.4% RPV lift. The question then becomes whether the margin loss from the discount is bigger than the RPV gain — RPV doesn't see gross margin.
AOV only counts buyers. If your test makes some visitors stop buying, those visitors disappear from the AOV calculation entirely — leaving you with a richer-looking buyer pool and a smaller revenue line. RPV catches what AOV hides because it divides by all visitors, not just converters.
Raising a free-shipping threshold (say €50 → €75) almost always lifts AOV — small baskets get pulled up — but suppresses conversion rate from price-sensitive visitors. Net RPV impact is usually small or negative. Lowering or removing the threshold typically lifts conversion rate and RPV but drops AOV.
Mid-market apparel typically lands in the €1.50–€3.50 RPV range, with stronger brands hitting €4–€6. The two levers are conversion rate (1.5–3.5% is typical) and AOV (€55–€95). Anything below €1 usually points to a traffic-quality or PDP-trust issue, not a checkout one.
AOV is useful for merchandising and unit economics — pricing, bundle design, shipping subsidy math, contribution margin per order. It's a diagnostic, not a scoreboard. Reporting both is fine; ranking tests by AOV alone is where teams get into trouble.
Multiply your baseline RPV by the proposed lift, then by monthly sessions, then by 12, to get annualised incremental revenue. Compare against the implementation and risk cost. A 3% RPV lift on a store doing 80k sessions/month at €2.50 RPV is roughly €72k/year — usually worth shipping if confidence is real.
Yes. Mobile RPV usually trails desktop RPV by 30–50% in apparel and beauty, mostly via conversion rate. Paid-search RPV is typically higher than paid-social RPV because of intent. Aggregate RPV hides these gaps — segmenting is where the experiment ideas come from.
RPV has higher variance than conversion rate because order values are skewed, so tests need more traffic — typically 1.3–2× the sample you'd need for a pure conversion-rate test. Two-week minimum runtime is standard regardless, to cover weekday/weekend mix and catch novelty effects.
Get an AI expert review of your site
Paste your URL — Metricuno's AI runs the same heuristic checks a senior CRO consultant would, scoring your page and prioritising the fixes that'll move conversion fastest.